noirdoc
PII guard for Claude Code to keep client data out of context
Noirdoc is an open‑source utility that intercepts files before they are sent to Claude Code, replacing personal identifiers such as names, email addresses, and IBANs with placeholder tokens. The redaction occurs locally on the user’s machine, so the original values never appear in Claude’s context. After Claude generates a response, noirdoc can restore the original data in the user’s terminal, keeping the conversation transcript free of sensitive information.
The tool is targeted at developers and teams that integrate Claude Code or other large‑language‑model workflows into their pipelines and need to protect client data. It offers three entry points: a Claude Code plugin, a command‑line interface for any LLM workflow, and a hosted proxy that can be used with OpenAI, Anthropic, or Azure services. Configuration is based on glob patterns, allowing users to specify which files or paths should be pseudonymized automatically.
All detection and mapping are performed locally, with pseudonymized copies stored in a hidden cache directory and the reversible mapping never leaving the machine. The project is released under the MIT License and can be installed via pip, with optional detectors available through an extra “full” package.
Reviews
Loading reviews…
Similar apps

Security & Identity
QuiGuard
Selfhosted proxy; scrubs secrets from AI Agent tool calls.

Clipboard, Input & Automation
MetaVeils
Sanitize documents before pasting into AI chatbots

Password & Security
BlankOut
Redact your docs on-device before sharing to AI

AI Coding Agents
Claudoscope
Browse, search & track costs across Claude Code sessions

Security & Identity
Locki Security
Protect PII and sensitive information in any web app

Security & Identity
Presidio JS - Client-Side PII Scanner
Anonymize prompts privately and securely